A number of potential mediators underlying this renal proton secretion have been identified (3). These include apical ATP-dependent H⁺ pumps and H⁺-K⁺ATPases on epithelial cells in the distal nephron; in addition, proton secretion is coupled to bicarbonate reabsorption mediated by the exchanger AE1. Human genetic studies have recently identified essential elements for pH homeostasis. To date, three genes in which mutations cause distal renal tubular acidosis have been identified (4-6). One of these genes, AE1, is essential for normal renal bicarbonate reabsorption. The other two genes encode subunits of the vacuolar-type H⁺ ATPase expressed in the apical plasma membrane of type A intercalated cells (ICs) of the collecting duct (CD); this H⁺ pump mediates secretion of protons into tubular fluid and can support a 1,000-fold concentration gradient (7, 8). Vacuolar-type H⁺ ATPases are homologous to the ATP synthase of the inner mitochondrial membrane and are composed of at least 13 subunits (9). These multisubunit complexes are believed to mediate acidification of intracellular compartments in all eukaryotic cells, whereas, in a subset of tissues, including the renal collecting duct, they mediate proton export from cells at the plasma membrane (8, 9). Some H⁺ ATPase subunits show specific isoforms that are largely restricted in their cellular localization (10). For example, the B2-subunit is largely found in intracellular compartments (11-14). In contrast, the B1- and a4-subunits, the subunits found mutated in renal tubular acidosis, each show specificity for the plasma membrane H⁺ ATPase (6, 11-14), and patients with homozygous loss of function mutations show systemic acidosis due to an inability to produce an appropriately acidic urine (5, 6).

Karyotypically normal targeted clones were injected into blastocysts to obtain chimeric mice, which were crossed with C57B/6 females to achieve germ-line transmission of the targeted allele in heterozygous, and ultimately homozygous, animals. Mice were genotyped by PCR using primer pairs specific for the wild-type (WT) (intron/exon 9) and deleted alleles (Neo cassette) and genomic DNA as template. Atp6v1b1^-/- mice derived from two independent targeted ES cell clones were phenotypically indistinguishable. ES cell manipulations were performed by the Transgenic Mouse and Gene Targeting Resource of Yale University.

Animal Care and Diet. Mice were maintained on a standard rodent diet (Prolab RMH3000, LabDiet) with free access to tap water, unless otherwise indicated. For acid-loading studies, a solution of 1.5% (280 mM) NH₄Cl/1% sucrose was substituted for tap water. All procedures were approved by the Yale University Animal Care and Use Committee or the Massachusetts General Hospital Institutional Committee on Research Animal Use.

Western Blotting. Whole mouse kidneys were homogenized in buffer containing protease inhibitors and centrifuged for 15 min at 13,000 × g at 4°C to yield total kidney extract. SDS/PAGE (30 μg of total protein per lane) and immunoblotting using a rabbit H⁺ ATPase B1-subunit polyclonal antibody (1:1,000) (14) and horseradish peroxidase-conjugated donkey anti-rabbit IgG (1:5,000, Jackson ImmunoResearch) was performed as described (14).

Immunohistochemistry and Histology. After vascular rinsing with PBS, mouse kidneys were fixed with paraformaldehyde-lysine-periodate by intracardiac perfusion of 6-week-old mice, and cryosections (5 μm) were prepared and immunostained (14, 15). Primary antibodies were rabbit polyclonal anti-H⁺ ATPase B1-subunit (1:150) (13); rabbit affinity-purified anti-H⁺ ATPase B2-subunit (1:10) (14); chicken affinity-purified anti-H⁺ ATPase E-subunit (1:100) (16); rabbit affinity-purified anti-AE1 (1:300, Chemicon); rabbit anti-pendrin (1:300, provided by Peter Aronson, Yale University); goat affinity-purified anti-aquaporin 2 (1:100, Santa Cruz Biotechnology). Donkey secondary antibodies were anti-rabbit IgG-Cy3 (1:1,000), anti-goat IgG-Cy5 (1:100), and anti-chicken IgG-FITC (1:60, Jackson ImmunoResearch). Sections (at least two animals per genotype) were imaged by confocal microscopy. Formalin-fixed renal sections from 10-month-old mice were stained with haematoxylin and eosin or van Kossa stain for assessment of histology and calcium phosphate deposition.

Biochemical Analysis and Physical Parameters. Venous blood obtained from the retroorbital plexus of conscious mice (2-3 months old) was immediately analyzed for blood gases and pH; electrolytes and creatinine were measured in the Yale-New Haven Hospital Clinical Chemistry Laboratory. Arterial blood of anesthetized mice (≥5 months old) was analyzed for K⁺ levels by flame photometry (17). Urine samples from 3- to 4-month-old mice were collected on Parafilm, and pH was measured by using a glass pH electrode. Urinary calcium levels and osmolality were determined by routine methods. Twenty-four-hour urine output was collected under oil from pairs of 6- to 8-week-old mice housed in metabolic cages (Nalgene), after 3-day adaptation to the cage. Assessments of body weight and of femoral bone density by using a PIXImus Mouse Densitometer (GE Medical Systems) were made on 9-week-old mice. Comparisons of biochemical parameters between groups were made by using the Mann-Whitney Wilcoxon rank sum test for two independent variables. P values <0.05 were considered statistically significant.

Surgical Methods. Renal clearance studies of adult mice (≥5 months old) were conducted by modifying a previous protocol (17). Saline (140 mM NaCl/5 mM KCl) was infused at 0.15 ml/h per 10 g of body weight; after 1.5 h of equilibration, urine output was collected under equilibrated mineral oil for a 30-min control period followed by bolus i.v. injection of furosemide (2 mg/kg) and collection of urine for a second 30-min period. Blood samples (30 μl) were taken immediately before and 1 h after furosemide infusion. Urine electrolytes and pH were measured and compared between genotypes by ANOVA with repeated measures. Comparisons of values within the same genotype before and after the administration of furosemide were made by using a least significant difference test (css statistica software). P values <0.05 were considered statistically significant.

Targeted clones were injected into blastocysts to generate Atp6v1b1^+/- chimeric mice; chimeric males were, in turn, bred with C57B/6 females. Tail DNAs of the resulting agouti offspring were screened by PCR (see Methods) to identify germ-line transmission of the mutant allele, and lines established from both ES clones transmitted the disrupted allele. Intercrossing of Atp6v1b1^+/- mice yielded offspring of all three genotypes (Fig. 1d) in expected Mendelian ratios; there was no evidence for decreased fetal or neonatal viability of Atp6v1b1^-/- mice. Atp6v1b1^-/- mice raised on a standard diet were fully fertile and showed normal growth and behavior when followed up to 1 year of age.

Western blotting showed that B1-subunit expression was absent in whole kidney homogenates of Atp6v1b1^-/- mice (Fig. 1e). Immunohistochemistry of kidneys of littermates showed that, whereas the B1-subunit was detected and indistinguishable in ICs of both Atp6v1b1^+/+ (Fig. 2a) and Atp6v1b1^+/- mice (data not shown), it was absent in Atp6v1b1^-/- mice (Fig. 2b). These results confirm that the targeted disruption eliminates production of the normal B1-subunit.

Cellular Organization of the Collecting Duct in Atp6v1b1^-/- Mice. Renal histology of 10-month-old Atp6v1b1^-/- mice appeared grossly normal with no evidence of nephrocalcinosis (data not shown). To assess renal CD epithelial cell subtype differentiation, we examined the expression of proteins that are normally found in specific cell types. The collecting duct is normally composed of type A ICs (A-ICs), which mediate proton secretion and bicarbonate reabsorption, type B ICs (B-ICs), which mediate bicarbonate secretion, and principal cells, which mediate electrogenic Na⁺ reabsorption. These cells are marked by anion exchanger 1 (AE1) on the basolateral membrane, pendrin on the apical membrane, and aquaporin 2 on the apical membrane, respectively. Each of these proteins was present and targeted to its normal subcellular location in kidneys of the Atp6v1b1^-/- mice (Fig. 2 c and d). These results suggest that differentiation of the collecting duct epithelia persists in Atp6v1b1^-/- mice.

Biochemical Parameters on a Standard Diet. The cardinal feature of humans with homozygous loss of ATP6V1B1 is systemic acidosis with an inappropriately alkaline urine. Among mice fed a standard rodent diet, there was no significant difference in venous pH or plasma bicarbonate level between Atp6v1b1^+/+ and Atp6v1b1^-/- mice (Fig. 3 a and b and Table 1). Nonetheless, urine pH was significantly higher among Atp6v1b1^-/- mice compared to Atp6v1b1^+/+ controls (pH 7.2 vs. 6.1, respectively, P = 0.0001; Fig. 3c and Table 1), indicating that the B1-subunit normally contributes to the acidic urinary pH.

Other plasma electrolytes showed no significant difference between Atp6v1b1^+/+ and Atp6v1b1^-/- mice, although Atp6v1b1^-/- mice showed a trend toward lower potassium levels; similarly, there was no significant difference between genotypes in mean arterial blood pressure, hematocrit, and glomerular filtration rate (Table 1). Compared to WT, there was reduced urine osmolality in Atp6v1b1^-/- mice (P = 0.03) and a trend toward greater 24-h urine output (Table 1). In contrast to the marked hypercalciuria seen in humans, urinary Ca²⁺/Cr ratios were not elevated, and were in fact significantly reduced in Atp6v1b1^-/- mice compared to WT (P = 0.002; Table 1). Bone densities of Atp6v1b1^+/+ and Atp6v1b1^-/- mice were not significantly different (Table 1).

Impaired Handling of an Acid Load in Atp6v1b1^-/- Mice. The difference in typical diets of mice and humans might account for the absence of systemic acidosis in Atp6v1b1^-/- mice. To determine whether the mutant mice can secrete an acid load, blood and urinary measurements were compared in Atp6v1b1^+/+ and Atp6v1b1^-/- mice after acid loading by administration of 1.5% (280 mM) NH₄Cl/1% sucrose in the drinking water. Although this acid load induced a significant reduction in systemic pH in both genotypes, acidosis was significantly more severe in Atp6v1b1^-/- mice (systemic pH 7.22 in WT vs. 7.13 in Atp6v1b1^-/-; P = 0.036; Fig. 3a). This increased acidosis in Atp6v1b1^-/- mice was accompanied by a significant reduction in venous plasma bicarbonate level in Atp6v1b1^-/- mice (19 vs. 16 mmol/liter; P = 0.036; Fig. 3b). This increased systemic acidosis is accompanied by a failure of Atp6v1b1^-/- mice to appropriately acidify the urine: despite having a lower systemic pH, these mice have a markedly more alkaline urine (pH 5.7 in WT vs. 6.4 in Atp6v1b1^-/-, P = 0.0001; Fig. 3c), corresponding to an 80% reduction in free proton concentration. This finding demonstrates that the B1-subunit is essential for achieving maximal urinary acidification. Urinary and venous pH levels at baseline and following acid challenge were not different between Atp6v1b1^+/- and Atp6v1b1^+/+ littermates (data not shown).

Electrogenic H⁺ Secretion in the CD Depends on the B1-Subunit. The above experiment provoked H⁺ secretion by increasing the blood concentration of H⁺. One can also provoke H⁺ secretion acutely by increasing the lumen-negative potential of the collecting duct, which increases the electrical driving force for H⁺ secretion. This increase was achieved by i.v. administration of furosemide to anesthetized Atp6v1b1^+/+ and Atp6v1b1^-/- mice (see Methods). Furosemide inhibits NaCl reabsorption in the thick ascending limb of Henle's loop, resulting in a marked natriuresis, which induces a compensatory increase in electrogenic Na⁺ reabsorption in the CD via the epithelial Na⁺ channel ENaC. This reabsorption increases the lumen-negative potential in the CD, thereby increasing the electrogenic driving force for H⁺ secretion (20).

Furosemide induced marked natriuresis in both genotypes; although there was slightly less natriuresis in the Atp6v1b1^-/- mice, this difference was not significant (Fig. 3d). Consistent with an increase in electrogenic driving force in the CD, furosemide induced marked and indistinguishable kaliuresis in both genotypes (Fig. 3e). In contrast, whereas furosemide induced a marked increase in urinary acidification in WT mice (10-fold increase in H⁺ concentration), there was no significant acidification in Atp6v1b1^-/- mice (Fig. 3f). This finding demonstrates that the normal urinary acidification induced by a lumennegative potential depends on the presence of functional B1-subunit.

Impaired Proton Extrusion from Intercalated Cells After Intracellular Acidification. To directly examine H⁺ efflux from individual ICs, cortical CD fragments were isolated. The ability of single ICs within these fragments to extrude H⁺ was measured after acute intracellular acidification induced by exposure to 20 mM NH₄Cl (Fig. 4 a and b). H⁺ extrusion was assessed by monitoring recovery of pH_i with the pH-sensitive dye BCECF (see Methods) in the presence and absence of 100 nM concanamycin (CON), a specific inhibitor of the vacuolar-type H⁺ ATPases (21), which include the apical H⁺ ATPase. In WT mice, mean pH_i recovery rate was reduced ≈70% by CON (Fig. 4c; 0.035 ± 0.001 vs. 0.009 ± 0.001 pH units/min), suggesting that this fraction of H⁺ efflux is mediated by the activity of plasma membrane H⁺ ATPases. The rate of pH_i recovery in Atp6v1b1^-/- ICs (0.011 ± 0.001 pH units/min) was markedly reduced compared to WT (Fig. 4c, mean rate of pH recovery reduced by 70%; P < 0.001). In contrast to WT, there was no further reduction in mean recovery of Atp6v1b1^-/- cells after addition of CON (0.009 ± 0.001 pH units/min; Fig. 4c). These results demonstrate that normal H⁺ excretion after acute intracellular acidification requires H⁺ ATPases containing the B1-subunit.

Localization of the H⁺ATPase B2-Subunit Is Altered in Atp6v1b1^-/- Medulla. In addition to the B1 isoform, expression of the alternative B subunit isoform (B2) has also been detected in type A ICs of both rat and mouse, predominantly in a vacuolar pattern but also on the apical membrane (14). To determine whether expression and/or localization of the alternative B2-subunit of the H⁺ ATPase is modified in Atp6v1b1^-/- kidney, we performed immunohistochemistry using a B2-specific antibody (Fig. 5). Apical staining was rare in the cortex and outer stripe of Atp6v1b1^+/+, Atp6v1b1^+/-, and Atp6v1b1^-/- outer medulla (data not shown). In Atp6v1b1^+/+ and Atp6v1b1^+/- mice, apical B2-subunit staining was found in a small minority of cells in the inner stripe (data not shown) and in 14% cells from the inner medulla (IM) (Fig. 5a; 1,332 cells counted from three animals). In contrast, the proportion of apically staining cells in the IM increased to 74% in Atp6v1b1^-/- mice (Fig. 5b; 704 cells counted from two animals).

Although human dRTA patients harboring homozygous B1-subunit mutations typically present as infants with spontaneous metabolic acidosis and failure to thrive (5), Atp6v1b1^-/- mice raised on a standard rodent diet were healthy, grew normally, and did not develop metabolic acidosis. This phenotypic discrepancy is likely related to dietary differences. Standard rodent chow has been shown to provide a large net dietary alkali load (23), whereas the typical Western human diet, with its higher protein content, provides a net acid load (1). Because studies in both humans (24) and rats (25) have shown a positive correlation between increasing dietary protein intake and net acid excretion, proton secretion mediated by the B1-subunit may become essential for the maintenance of systemic pH homeostasis only in the context of a net dietary acid load. Nevertheless, the significant difference in urine pH indicates that the B1-subunit does play a role in mice on a standard diet.

A second difference between the phenotype of humans and mice deficient for Atp6v1b1 is the uniform hypercalciuria in humans (5), which is absent in the mouse. Atp6v1b1^-/- mice had normal bone density, did not develop nephrocalcinosis, and, in fact, displayed significantly lower urine calcium excretion compared to WT. The absence of skeletal abnormalities in Atp6v1b1^-/- mice supports the hypothesis that hypercalciuria in humans is attributable to dissolution of bone by the systemic acidosis, rather than a direct role of Atp6v1b1 in bone metabolism. Our findings are consistent with the observation that B2, rather than B1, is the B-subunit expressed in osteoclasts (26). Finally, study of Atp6v1b1^-/- mice has yet to demonstrate an essential role for the B1-subunit in non-renal tissues known to express this protein in WT mice, such as male reproductive tract and inner ear. Although it has been proposed that H⁺ ATPase activity in the male reproductive tract plays a role in fertility (16), male Atp6v1b1^-/- mice were fertile. Although sensorineural hearing loss cosegregates with dRTA in kindreds harboring B1-subunit mutations (5), inner ear histology, auditory brainstem responses, and endocochlear potential were normal in Atp6v1b1^-/- mice (27).

In the absence of the B1-subunit, we found a compensatory increase in apical H⁺ ATPase complexes containing the B2-subunit in IM. Although the functional significance of this increased apical B2-subunit expression is uncertain, these results raise the possibility that, under some circumstances, the B2-subunit may provide partial functional compensation for the loss of the B1-subunit. Because Atp6v1b1^-/- mice did achieve some increase in urinary acidification after oral acid challenge, apical B2-subunit-containing H⁺ ATPase complexes could indeed be contributing to proton secretion in this setting, although not at a level sufficient to maintain normal pH homeostasis.

Whether the B1-subunit contributes to H⁺ ATPases that mediate acidification of intracellular compartments such as endosomes and lysosomes has remained a matter of debate. Because Atp6v1b1^-/- mice do not display evidence of generalized renal tubular dysfunction, our results suggest that the B1-subunit is unlikely to be required for endocytic acidification processes in renal epithelia.